Radio frequency (RF) and microwave devices are typically characterized by an N by N matrix of scattering parameters (S-parameters) where N is the number of ports of the RF or microwave device. Each S-parameter, denoted Sxy, is the ratio of the output signal at port x to the incident signal at port y when no other signals are incident on the RF or microwave device. For example, the ratio of the output signal at port 2 to the incident signal at port 1 is designated as S21.
S-parameters are typically measured by a S-parameter measurement device such as a network analyzer. Most network analyzers only have two measurement ports, although there are network analyzers available with four or more measurement ports. To accommodate a multi-port device under test (DUT) with a two-port network analyzer, the two-port network analyzer is often connected to a programmable switch matrix that contains at least as many ports as the number of ports on the DUT. The network analyzer measures the S-parameters of the DUT by configuring the switch matrix to couple two DUT ports to the two measurement ports. The remaining DUT ports are terminated in the switch matrix.
Regardless of the type of S-parameter measurement device used, and whether a switch matrix is used, the S-parameter measurement system will inevitably include hardware imperfections that can produce measurement errors if they are not accounted for in the measurements. The errors can either be systematic or random. Systematic errors are caused by imperfections in the test equipment and test setup. For example, systematic errors can result from directivity effects in the couplers, cable losses and mismatches between the S-parameter test system and the DUT. These errors can typically be characterized through calibration and mathematically removed during the measurement process. For a standard two-port network analyzer measurement, there is a generally accepted calibration model that is used to remove the effects of the following twelve systematic errors:    forward directivity(EDF)    forward crossisolation(EXF)    forward sourcematch(ESF)    forward reflectiontracking(ERF)    forward load match(ELF)    forward transmissiontracking(ETF)    reversedirectivity(EDR)    reversecrossisolation(EXR)    reversesourcematch(ESR)    reversereflectiontracking(ERR)    reverseloadmatch(ELR)    reversetransmissiontracking(ETR)
In contrast, random errors vary randomly as a function of time and thus cannot be removed by calibration. The main contributors to random errors are instrument noise, switch repeatability and connector repeatability.
Even if the system is calibrated, all of the errors may not be removed because after calibration the system may be subject to a number of changes that cause calibration drift. Specifically, changes in temperature, changes due to switching in the RF or microwave path and movement in the cables. When electromechanical RF or microwave switches are used on the test system, their RF or microwave characteristics change when they are actuated. Cable movement also affects the system characteristics and the higher the frequency the more significant the affects. Accordingly, if the calibration drift is not accounted for then the S-parameter measurements will lose their accuracy over time.
Calibration drift is typically minimized or eliminated by one of the following methods: use of silicon dioxide (SiO2) test cables to minimize test system loss and improve phase stability versus temperature; use of high repeatability switches; or, use of reference cables to compensate the effect of systematic changes. Each of these techniques, however, is associated with a particular limitation. Specifically, high repeatability switches are expensive and degrade substantially in repeatability over life. Furthermore, while silicon dioxide cables are very stable over temperature, they do exhibit changes as a function of temperature. In addition, the effectiveness of these methods is significantly reduced as the test frequencies reach the Ka band or when the test system losses increase, for example, when long test cables are used to connect the test system to the DUT.
However, there are many applications that require long cables between the test system and the DUT. For example, it may be necessary to put the DUT into a specific environment such as inside a vacuum chamber, a thermal chamber or an anechoic chamber. It also may be necessary to test the DUT “in situ” such as mechanically integrated into a large assembly. Accordingly, there is a need for a system that can dynamically correct S-parameter measurements to account for calibration drift.